Instagram Signals Could Flag Loneliness

A study of 415 undergraduates (including private accounts) used machine learning on photos, captions, likes and timing to predict self‑reported loneliness and to distinguish 'emotional' vs 'social' loneliness by different visual/linguistic markers. The authors suggest platforms could use such signals to trigger tailored outreach (e.g., nudges to connect or crisis resources). — If platforms operationalize these signals, it raises tradeoffs between potentially helpful early mental‑health interventions and mass privacy‑sensitive behavioral surveillance and automated gatekeeping.